Using values defined in the BBOB document

[JJ]

According to the "General Setup" section of the BBOB benchmark paper, the range is supposed to be between -5 and 5. Defined also a couple of constants so that it's clear what is meant.

[jonathanBieler]

Actually I think -4,4 was correct, no @JJ ? Or it's in another version of the doc ?

Search Space

All functions are defined and can be evaluated over RD, while the actual search domain is given as [−5,5]D.

Location of the optimal xopt and of fopt = f(xopt)

All functions have their global optimum in [−5,5]D. The majority of functions has the global optimum in [−4,4]D and for many of them xopt is drawn uniformly from this compact. The value for fopt is drawn from a Cauchy distributed random variable, with zero median and with roughly 50% of the values between -100 and 100. The value is rounded after two decimal places and set to ±1000 if its absolute value exceeds 1000. In the function definitions a transformed variable vector z is often used instead of the argument x. The vector z has its optimum in zopt = 0, if not stated otherwise.

https://inria.hal.science/inria-00362633/document/

[JJ]

From what I read, it looks like the search domain is different from the place where the actual optimum usually is. What I understand is that the optimum moves around within that subset. You can (slightly) cheat and use that as the search domain, but IIRC, BBOB benchmarks would require [-5,5]D as the search domain, as said in the first paragraph.